Complementaries in the Age of the App
11 December 2012 at 12:32 am Peter G. Klein 3 comments
| Peter Klein |
Josh Gans asks if “we have yet evolved to the right set of institutions in the app economy,” comparing contracts between app developers and distributors/publishers to those between book authors and publishers. He also notes, correctly I think, that app development may have more to do with signaling programming skill than making money from selling the app. Still, there are important contractual issues to be sorted out in the age of the app.
More generally, Josh’s post highlights the need for organizational scholars to think more broadly about the complementarities between technology, organization, and strategy. Milgrom and Roberts (1990, 1995) are the pioneers here, but there management literatures on modularity and other aspects of fit among organizational attributes are relevant too. (Here’s an example from outside the tech sector.) Milgrom and Roberts put it this way:
[C]hange in a system marked by strong and widespread complementarities may be difficult and . . . centrally directed change may be important for altering systems. Changing only a few of the system elements at a time to their optimal values may not come at all close to achieving all the benefits that are available through a fully coordinated move, and may even have negative payoffs. Of course, if those making the choices fail to recognize all the dimensions across which the complementarities operate, then they may fail to make the full range of necessary adaptations, with unfortunate results. At the same time, coordinating the general direction of a move may substantially ease the coordination problem while still retaining most of the potential benefits of change. Moreover, the systematic errors associated with centrally directed change are less costly than similarly large but uncoordinated errors of independently operating units.
In other words, when a system is characterized by strong complementarities, the diffusion and evolution of business practices requires simultaneous, coordinated changes among all complementary features within the system — technology, organizational form, strategy, and perhaps other elements as well. When simultaneous or coordinated changes occur within strongly complementary systems, business practices like contractual form will also tend to evolve, and to do so rapidly. By contrast, when simultaneous or coordinated changes within systems characterized by strong complementarities do not occur, organizational change will tend to be slow or uneven.
The rapid growth of the app economy might seem an exception to these principles, as the app market has exploded without (it appears) complementary changes in the contractual and organizational aspects of app production. As noted above, this may be because app design performs a signaling role independent of its ability to generate profits. If this becomes less important over time — perhaps because clever programmers find more effective ways to signal ability — then getting the compensation system right will be critical to ensure the success of this particular business model.
Entry filed under: - Klein -, Innovation, Institutions, New Institutional Economics, Strategic Management.
1.
stevepostrel | 13 December 2012 at 12:50 am
Funny you should mention Milgrom-Roberts and complementarities. I just gave a presentation about some of the conceptual problems with their definition of complementarities and some ideas for fixing it.
2.
Peter Klein | 13 December 2012 at 1:00 am
Great. What did you say? Got slides? Or a write-up?
3.
stevepostrel | 13 December 2012 at 7:31 pm
I have slides, but that was a very-first tryout and some of them are pretty crappy. Working on the paper.
Main idea: If we care about complementarity because we think getting the mixture of practices right is important, then the M-R definition creates confounding problems with returns to scale. This confound potentially generates type I and type II error problems in identifying complementarity. The problem gets even worse when the “inputs” we study are opportunity observables (e.g. presence or absence of incentive pay, cross-functional teams, worker autonomy, training, IT) that contain unknown mixtures of the true but possibly unknown factors actually driving outcomes.
I hope that’s not too clear–I need to maintain a little bit of mystery.